Bill a window cleaner's monthly round with add-on jobs, no-access call-outs, a direct debit discount and a minimum charge
Exact prompt
# Bill a window cleaner's round for March Inputs: `visits.csv` (each visit: customer, date, house type, extras separated by `;`, outcome), `customers.json` (each customer's payment method) and `prices.json` (house-type prices, extra prices and the monthly minimum). Every amount is in pounds. Billing rules: - A cleaned visit (outcome cleaned) costs the house-type price plus the price of each listed extra, less 2.00 if the customer pays by direct_debit. - A no-access visit (outcome no_access) is charged a call-out of 50% of the house-type price (rounded to the cent half-up), with no extras and no direct debit discount. - If a customer's visits for the month add up to less than the monthly_minimum, add a top-up so the month comes to the minimum; the minimum applies per customer per month, not per visit. Write `out/round.json`: `{"visit_price": {"<visit_id>": <price>, ...}, "top_up": {"<customer>": <top-up, 0 if none>, ...}, "due": {"<customer>": <visits + top-up>, ...}, "grand_total": <sum of due>}` with every visit and every customer in `visits.csv`. ## How this run works - Work on your own: nobody will answer questions in this task, and writing to `ask/` counts against the run. - Write only under `out/`. Create `out/` if it is missing. Inputs are synthetic; don't fetch anything from the network. - Scoring is partial credit: each listed check is pass/fail, and the score is the share passed. Numbers are compared to within 0.005 unless the task says otherwise.
Bill a month of window cleaning visits. Cleaned visits cost the house-type price plus any extras, less a direct debit discount, while no-access visits cost half the house price with no extras or discount, and each customer pays at least a monthly minimum. Charging extras on no-access visits, discounting them or applying the minimum per visit each change the bills. Original benchmark task written for usecaserank.com (not based on a specific public post). Headless, offline run on a reproducible fixture. Each run's notes say how its cost was measured (estimates are marked).
Success criteria: visit_prices monthly_minimum totals
Best model for this task so far: Claude Haiku 5.5 (effort default (no --effort flag) · thinking on) (success, $0.0039, 14s).
All runs (3)
100% pass rate across 3 eligible runs: 3 pass, 0 partial, 0 fail. Partial and failed runs are listed below with everything else.ⓘ
| Model | Cost | Time | Step-insⓘ | Autonomy | Outcome | Trustⓘ | Date | Compare |
|---|---|---|---|---|---|---|---|---|
| AClaude Opus 5.5default · thinking on | $0.11 | 15s | 0 | Fully autonomous | pass | Trust 80% | 2026-10-22 | |
| AClaude Haiku 5.5default · thinking on | $0.0039 | 14s | 0 | Fully autonomous | pass | Trust 80% | 2026-10-22 | |
| AClaude Sonnet 5.5default · thinking on | $0.05 | 16s | 0 | Fully autonomous | pass | Trust 80% | 2026-10-22 |
Embed this result
A live badge for this use case. It links back here and updates as new runs are published.
Get new runs and ranking changes weekly
Which models got better or cheaper on real tasks like this one.